List your company
ResearchFormerly WizWand

Semantic Segmentation on COCO-Object with background class

67.9mIoUCoCo-SAM3
4.04420.62237.253.778Dec 1, 2022Jun 25, 2023Jan 17, 2024Aug 10, 2024Mar 4, 2025Sep 26, 2025Apr 21, 2026
Updated 6mo ago

Evaluation results

MethodLinks
CoCo-SAM3Paradigm=Training-free SAM32026.0467.9
SAM3Paradigm=Training-free SAM32026.0465.4
CorrCLIPParadigm=Training-free CLIP-VFM2026.0449.4
ReMEParadigm=Training-free CLIP-VFM2026.0448.2
Talk2DINOParadigm=Training-based2026.0445.1
CLIPtraseParadigm=Training-free CLIP-Only2026.0443.6
TridentParadigm=Training-free CLIP-VFM2026.0441.1
RF-CLIPParadigm=Training-free CLIP-Only2026.0439.1
SFPParadigm=Training-free CLIP-Only2026.0437.9
CASSParadigm=Training-free CLIP-VFM2026.0437.8
ProxyCLIPParadigm=Training-free CLIP-VFM2026.0437.4
FreeDAParadigm=Training-free CLIP-VFM2026.0436.3
CLIP-DINOiserParadigm=Training-based2026.0434.8
NACLIPParadigm=Training-free CLIP-Only2026.0433.2
ClearCLIPParadigm=Training-free CLIP-Only2026.0433
TCLEvaluation Mode=Zero-shot2022.1231.6
COSMOS w/ SCLIPData Size=30M, Zero-shot=true, Vision Encoder Architecture=ViT-B/162024.1231.3
SCLIPData Size=400M, Zero-shot=true, Vision Encoder Architecture=ViT-B/162024.1230.5
SCLIPParadigm=Training-free CLIP-Only2026.0430.5
TCLParadigm=Training-based2026.0430.4
GroupViT (RedCaps)Training Data=RedCaps + CC12M2022.1227.5
GroupViTParadigm=Training-based2026.0427.5
GroupViT (YFCC)Training Data=YFCC + CC12M2022.1224.3
MaskCLIPRefinement=With heuristic refinement2022.1220.6
MaskCLIPData Size=400M, Zero-shot=true, Vision Encoder Architecture=ViT-B/162024.1220.6
ReCoEvaluation Mode=Zero-shot2022.1215.7
MaskCLIP (Baseline)Refinement=None2022.1215.5
CaRParadigm=Training-free CLIP-Only2026.0415.4
COSMOSData Size=30M, Zero-shot=true, Vision Encoder Architecture=ViT-B/162024.1215.2
OpenCLIPData Size=1B, Zero-shot=true, Vision Encoder Architecture=ViT-B/162024.1211.4
OpenCLIPData Size=2B, Zero-shot=true, Vision Encoder Architecture=ViT-B/162024.129.9
OpenCLIPData Size=400M, Zero-shot=true, Vision Encoder Architecture=ViT-B/162024.128.7
CLIPData Size=400M, Zero-shot=true, Vision Encoder Architecture=ViT-B/162024.128.1
CLIPParadigm=Training-free CLIP-Only2026.046.5